@nejch had a quick question. I've written a functional test for this functionality where I'm creating some entities (projects, groups, snippets, users, etc.) and then checking whether the statistics returned match or not (number of projects created = number from statistics).
However, this test keeps failing because there are a number of projects created in other functional tests that were not cleaned up, and hence those numbers are interfering with the test. Similarly for other entities.

I have two options here: either I go around modifying those tests to clean up any entities they create, or I do a count of created (and undeleted) entities and just modify my test accordingly.
I would definitely prefer the first one, since it makes my test independent of others -- what do you think?

@nejch had a quick question. I've written a functional test for this functionality where I'm creating some entities (projects, groups, snippets, users, etc.) and then checking whether the statistics returned match or not (number of projects created = number from statistics). However, this test keeps failing because there are a number of projects created in other functional tests that were not cleaned up, and hence those numbers are interfering with the test. Similarly for other entities.

I have two options here: either I go around modifying those tests to clean up any entities they create, or I do a count of created (and undeleted) entities and just modify my test accordingly. I would definitely prefer the first one, since it makes my test independent of others -- what do you think?

Jumping in if I may. The fixture reset_gitlab can do the trick.

Jumping in if I may. The fixture reset_gitlab can do the trick.

Just what I was looking for, thanks!

Codecov Report

Merging #2347 (6619835) into main (f04e8ba) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##             main    #2347   +/-   ##
=======================================
  Coverage   95.78%   95.79%           
=======================================
  Files          79       79           
  Lines        5249     5258    +9     
=======================================
+ Hits         5028     5037    +9     
  Misses        221      221           

Flags with carried forward coverage won't be shown. Click here to find out more.

Sorry @Shreya-7 I was away on a long weekend and catching up a bit!

I have two options here: either I go around modifying those tests to clean up any entities they create, or I do a count of created (and undeleted) entities and just modify my test accordingly.
I would definitely prefer the first one, since it makes my test independent of others -- what do you think?

I would actually keep it even simpler than that, and only check that the attributes in the response can be successfully cast to an int but not actually assert on their values, that way we know the server response makes sense.

GitLab does a lot of creation and cleanup tasks asynchronously and I would not rely on absolute values here. We've actually tried to move away from that in some other tests - see #2118 for context.

That said, I see you've done a lot of work on getting reset_gitlab to behave. So we can keep that as it likely helps the testing in any case (in a separate commit maybe). WDYT?

I would actually keep it even simpler than that, and only check that the attributes in the response can be successfully cast to an int but not actually assert on their values, that way we know the server response makes sense.

This makes sense, let me try that!

That said, I see you've done a lot of work on getting reset_gitlab to behave. So we can keep that as it likely helps the testing in any case (in a separate commit maybe). WDYT?

So should I open a new PR for this fix? I'm fine with it going in with this one, since my test does use this function